回顾Seata基于AT模式控制事务的流程

一 概述

        Seata 是一款开源的分布式事务解决方案，致力于提供高性能和简单易用的分布式事务服务。Seata 将为用户提供了 AT、TCC、SAGA 和 XA 事务模式，为用户打造一站式的分布式解决方案。

二 基于AT模式的事务控制流程

        1. 首先每个微服务都必须创建undo_log表；

        2. 安装事务协调器（TC）—— seata-server: 
Releases · seata/seata · GitHubhttps://github.com/seata/seata/releases

 三 整合SpringBoot

        添加依赖

<dependency>
    <groupId>com.alibaba.cloud</groupId>
    <artifactId>spring-cloud-starter-alibaba-seata</artifactId>
</dependency>

四 配置事务协调器

       在配置文件中存在file.conf和registry.conf两个配置文件。服务注册名称为Service-addr

       registory.conf中的注册中心配置，默认在file.conf中保存配置

registry {
  # file 、nacos 、eureka、redis、zk、consul、etcd3、sofa
  type = "file"

  nacos {
    application = "seata-server"
    serverAddr = "127.0.0.1:8848"
    group = "SEATA_GROUP"
    namespace = ""
    cluster = "default"
    username = ""
    password = ""
  }
  eureka {
    serviceUrl = "http://localhost:8761/eureka"
    application = "default"
    weight = "1"
  }
  redis {
    serverAddr = "localhost:6379"
    db = 0
    password = ""
    cluster = "default"
    timeout = 0
  }
  zk {
    cluster = "default"
    serverAddr = "127.0.0.1:2181"
    sessionTimeout = 6000
    connectTimeout = 2000
    username = ""
    password = ""
  }
  consul {
    cluster = "default"
    serverAddr = "127.0.0.1:8500"
    aclToken = ""
  }
  etcd3 {
    cluster = "default"
    serverAddr = "http://localhost:2379"
  }
  sofa {
    serverAddr = "127.0.0.1:9603"
    application = "default"
    region = "DEFAULT_ZONE"
    datacenter = "DefaultDataCenter"
    cluster = "default"
    group = "SEATA_GROUP"
    addressWaitTime = "3000"
  }
  file {
    name = "file.conf"
  }
}

        registry.conf中的配置中心配置，默认在file.conf中保存配置配置

config {
  # file、nacos 、apollo、zk、consul、etcd3
  type = "file"

  nacos {
    serverAddr = "127.0.0.1:8848"
    namespace = ""
    group = "SEATA_GROUP"
    username = ""
    password = ""
    dataId = "seataServer.properties"
  }
  consul {
    serverAddr = "127.0.0.1:8500"
    aclToken = ""
  }
  apollo {
    appId = "seata-server"
    ## apolloConfigService will cover apolloMeta
    apolloMeta = "http://192.168.1.204:8801"
    apolloConfigService = "http://192.168.1.204:8080"
    namespace = "application"
    apolloAccesskeySecret = ""
    cluster = "seata"
  }
  zk {
    serverAddr = "127.0.0.1:2181"
    sessionTimeout = 6000
    connectTimeout = 2000
    username = ""
    password = ""
    nodePath = "/seata/seata.properties"
  }
  etcd3 {
    serverAddr = "http://localhost:2379"
  }
  file {
    name = "file.conf"
  }
}

        上述配置可以根据修改type值，并对值相应的配置信息进行配置处理即可。

        file.conf配置信息

## transaction log store, only used in seata-server 事务日志存放的地方
store {
  ## store mode: file、db、redis
  mode = "file"
  ## rsa decryption public key
  publicKey = ""
  ## file store property
  file {
    ## store location dir
    dir = "sessionStore"
    # branch session size , if exceeded first try compress lockkey, still exceeded throws exceptions
    maxBranchSessionSize = 16384
    # globe session size , if exceeded throws exceptions
    maxGlobalSessionSize = 512
    # file buffer size , if exceeded allocate new buffer
    fileWriteBufferCacheSize = 16384
    # when recover batch read size
    sessionReloadReadSize = 100
    # async, sync
    flushDiskMode = async
  }

  ## database store property
  db {
    ## the implement of javax.sql.DataSource, such as DruidDataSource(druid)/BasicDataSource(dbcp)/HikariDataSource(hikari) etc.
    datasource = "druid"
    ## mysql/oracle/postgresql/h2/oceanbase etc.
    dbType = "mysql"
    driverClassName = "com.mysql.jdbc.Driver"
    ## if using mysql to store the data, recommend add rewriteBatchedStatements=true in jdbc connection param
    url = "jdbc:mysql://127.0.0.1:3306/seata?rewriteBatchedStatements=true"
    user = "mysql"
    password = "mysql"
    minConn = 5
    maxConn = 100
    globalTable = "global_table"
    branchTable = "branch_table"
    lockTable = "lock_table"
    queryLimit = 100
    maxWait = 5000
  }

  ## redis store property
  redis {
    ## redis mode: single、sentinel
    mode = "single"
    ## single mode property
    single {
      host = "127.0.0.1"
      port = "6379"
    }
    ## sentinel mode property
    sentinel {
      masterName = ""
      ## such as "10.28.235.65:26379,10.28.235.65:26380,10.28.235.65:26381"
      sentinelHosts = ""
    }
    password = ""
    database = "0"
    minConn = 1
    maxConn = 10
    maxTotal = 100
    queryLimit = 100
  }
}

五 通过注解使用seata

        分布式事务添加入口标注：@GlobalTransactional

        每个子事务添加入口标注：@Transational
        当使用到数据库的时候则需要配置代理数据源，如果没有则事务无法完成回滚：

@Configuration
public class MySeataConfig {

    @Autowired
    DataSourceProperties dataSourceProperties;

    @Bean
    public DataSource dataSource(DataSourceProperties dataSourceProperties){
        //properties.initializeDataSourceBuilder().type(type).build();
        HikariDataSource dataSource = dataSourceProperties.initializeDataSourceBuilder().type(HikariDataSource.class).build();
        if (StringUtils.hasText(dataSourceProperties.getName())) {
            dataSource.setPoolName(dataSourceProperties.getName());
        }
        return new DataSourceProxy(dataSource);
    }
}

       配置服务：

vgroup_mapping.{application.name}-fescar-service-group = "default"

        注：SpringBoot默认使用的数据源为Hikari

        上述方式并不适用于高并发场景。

        参考文档：
Seata 快速开始https://seata.io/zh-cn/docs/user/quickstart.html